Video Temporal Grounding on ActivityNet (R1 Metrics)
40.6R1@0.5TaRO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| TaROSize=8B, Base Model=Qwen3-VL-8B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 40.6 | 59.7 | 23.9 | |
| TaROSize=7B, Base Model=Qwen2.5-VL-7B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 39.8 | 60.6 | 21.4 | |
| Time-R1Size=7B, Base Model=Qwen2.5-VL-7B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 39 | 58.6 | 21.4 | |
| TaROSize=3B, Base Model=Qwen2.5-VL-3B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 32.7 | 52.3 | 14.2 | |
| VideoChat-R1.5Size=7B, Base Model=Qwen2.5-VL-7B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 32.3 | 52.4 | 16.8 | |
| Qwen3-VL-8B-InstructSize=8B, Base Model=Qwen3-VL-8B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 31.7 | 45.9 | 19.8 | |
| Qwen3-VL-8B-ThinkSize=8B, Base Model=Qwen3-VL-8B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 30.5 | 44 | 19.2 | |
| HawkEyeSize=7B, Evaluation Protocol=Zero-shot2026.06 | 29.3 | 49.1 | 10.7 | |
| VTimeLLMSize=7B, Evaluation Protocol=Zero-shot2026.06 | 27.8 | 44 | 14.3 | |
| Time-R1Size=3B, Base Model=Qwen2.5-VL-3B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 26 | 46.2 | 11.4 | |
| UniTimeSize=7B, Base Model=Qwen2.5-VL-7B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 22.8 | — | 14.1 | |
| ChatVTGSize=7B, Evaluation Protocol=Zero-shot2026.06 | 22.5 | 40.7 | 9.4 | |
| TimeChatSize=7B, Evaluation Protocol=Zero-shot2026.06 | 20.2 | 36.2 | 9.5 | |
| TimeSuiteSize=7B, Evaluation Protocol=Zero-shot2026.06 | 16.6 | — | 9.28 | |
| Qwen2.5-VL-3B-InstructSize=3B, Base Model=Qwen2.5-VL-3B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 15.2 | 26.5 | 7.2 | |
| Qwen2.5-VL-7B-InstructSize=7B, Base Model=Qwen2.5-VL-7B-Instruct, Evaluation Protocol=Zero-shot2026.06 | 13.6 | 24.4 | 6.7 |